README.md

metadata

title: SIQ Proxy
emoji: 🛰️
colorFrom: indigo
colorTo: purple
sdk: docker
app_port: 7860
pinned: false

siq-proxy

Streaming OpenAI-compatible proxy for the SIQ-1-35B RunPod serverless endpoint, fronting the Pi Agent web app's "SIQ-1-35B (cloud)" model. Holds the RunPod API key server-side, adds CORS, applies SIQ-1 request shaping (thinking toggle, reasoning effort, max_tokens floor), and streams the vLLM worker's SSE through to the browser.

Endpoints

• GET /api/health → { ok, siq, model }
• GET /api/siq/v1/models
• POST /api/siq/v1/chat/completions (OpenAI, stream: true)

The web app points VITE_SIQ_API at https://<this-space>.hf.space/api/siq.

Secrets (Space settings → Variables and secrets)

• RUNPOD_API_KEY (required) — RunPod API key
• SIQ_EID (required) — RunPod serverless endpoint id
• SIQ_MODEL (default siq) — served-model-name on the worker
• SIQ_MINTOK (default 2048), SIQ_LIMIT_HOUR_IP (default 240)